New cloud and HPC infrastructure and the launch of the Melbourne Research Cloud Dashboard
This week Research Platform Services releases a number of major research computing infrastructure upgrades to the research community. These upgrades include the addition of almost 4 petabytes of compute storage, 1000 high performance computing (HPC) cores, and replacement of nearly 4000 research cloud cores and addition of new speciality big memory and VDI nodes, along with many background service improvements. This coincides with the launch of the University’s new Melbourne Research Cloud Dashboard offering an improved user experience when accessing cloud cores for research computing. Later in 2019, this Dashboard will be extended to cover all research computing assets.
These upgrades and improvements have been made in response to feedback from users across all academic divisions, with priority-setting and budgeting support provided to Research Platform Services through its close collaboration with the University’s Petascale Campus Initiative (PCI).
An upgraded research network with high core and high memory options, starting at 25Gb and ranging up to 100Gb, enables both traditional compute on a larger scale or memory-intensive data analysis, with all regular instances having a default 1:4 vcpu/memory ratio. Specifically, this means:
- there are now over 2,500 cores on the University’s Spartan HPC cluster
- our standard cloud service now has over 10,000 virtual cores with up to 8 cores and 32Gb memory per instance
- our premium cloud service has around 10,000 virtual cores on new hardware, and includes:
- high core count flavours of up to 36 cores per instance, with a 1 core to 9Gb memory ratio
- high memory flavours of up to 2Tb per instance, with a 1 core to 42Gb memory ratio
- general purpose GPUs to for parallel processing and for machine learning applications
- VDI GPUs for high-capacity virtual desktops.
Features of the new Melbourne Research Cloud Dashboard
This is the first stage in the roll-out of a new look research computing interface planned for 2019. From the dashboard, users have a streamlined and integrated allocation request process which incorporates the variety of cloud computing offerings announced today, giving direct access to all new features, including premium flavours and GPGPU and VDI nodes. From this one access point, users can now get specialised database systems, get applications on demand (for those who just want to spin up an RStudio or Jupyter instance), and select the size of their instance which best suits their needs.
Getting started with your Dashboard
Whether you are an experienced user or a novice, you can get started with the new dashboard today.
If you don't have an existing project, one will be created for you as soon as you login for the first time.
If you already have a research cloud project, you’ll be able to access it immediately from the new dashboard.